Using DQN for the gym lunar lander environment. Everything works but I made a few iffy design choices out of laziness.
pip install .
You may also need to run apt install python-opengl.
To be run at the root of the project.
- python3 src/lunar_lander_dqn/run.py --tune config/tune.yaml --num_episodes 1000
- python3 src/lunar_lander_dqn/run.py --train config/train.yaml --num_episodes 1000
- python3 src/lunar_lander_dqn/run.py --run model/model.pth --num_episodes 10
- Make mypy happy
- Add input protections from an ill formed training/tuning configuration file
- Set up CI pipeline for project (pylint, mypy, tests, etc)